In this exercise you will learn how add the values in two arrays and write the
results to another array in parallel using parallel_for.
Create a queue to enqueue your kernel function to, remember to handle errors.
Create buffers to manage the data of the two input arrays and output array.
Remember to ensure the range provided to the buffer is the size of the arrays.
Create accessors to each of the buffers within the command group function.
Now enqueue parallel kernel function by calling parallel_for on the handler.
range and id variantThis version of parallel_for takes a range specifying the number of
iterations of the kernel function to invoke and the kernel function itself must
take an id which represents the current iteration.
The id can be used in the accessor subscript operator to access or assign to
the corresponding element of data that the accessor represents.
nd_range and nd_item variantThis version of parallel_for takes an nd_range which is made up of two
ranges describing the global range and the local range (work-group size). The
kernel function must take an nd_item, which cannot be passed directly to the
subscript operator of an accessor. Instead, retrieve the id using the
get_global_id member function.
For DevCloud via JupiterLab follow these instructions.
For DPC++: instructions.
For AdaptiveCpp: instructions.